The Electronic Structure of Hydrogen Fluoride - ACS Publications

molecular orbitals(M0s) are expressed as linear combinations of all valence and inner shell atomic orbitals of the atoms, and all interactions of the ...
0 downloads 0 Views 492KB Size
THEELECTRONIC STRUCTURE OF HYDROGEN FLUORIDE

April 20, 1955

[COXTRIBUTION FROM THE

DEPARTMENT O F CHEMISTRY,

2107

UNIVERSITY OF ROCHESTER]

The Electronic Structure of Hydrogen Fluoride' BY A. B. F. DUNCAN RECEIVED JUNE 25, 1954 A self-consistent field calculation of the normal state of HF in the molecular orbital(M0) approximation is made. The molecular orbitals(M0s) are expressed as linear combinations of all valence and inner shell atomic orbitals of the atoms, and all interactions of the ten electrons are included. The MO functions are used to compute the dissociation energy and dipole moment of the molecule. An equivalent orbital representation is obtained by an orthonormal transformation of the MOs which shows definite directional features of the distribution of electron density. The latter is discussed briefly in connection with the dipole moment and with association in HF.

The structure of a number of simple molecules, including HF, has been treated recently by a method of equivalent orbitak2 I n HF, the lone pair electrons were assigned to three equivalent orbitals, which were represented by s - p hybridized functions directed in three azimuthal planes a t angles of 120' to each other. Orthogonality between the lone pair functions requires that the angle between the axis of a lone pair function and the bond axis should be greater than 90°, but the actual angle could not be determined. Consequently the angle between the lone pair axes is not known, and only tentative conclusions could be drawn concerning the electron density distribution in the lone pair region of the molecule. Equivalent orbitals are derived from molecular orbitals(MOs), by orthonormal transformation^,^ and if a good (or the best) set of MOs is known, the lone pair distribution, and its contribution to the dipole moment of HF can be discussed more precisely. The purpose of the present calculations was to find such MOs by a self-consistent-field (SCF) procedure. I n this calculation, all ten electrons have been assigned to MOs and all interactions have been included. Thus the importance of inner-shell-outer-shell mixing can be assessed, as has been done in a recent calculation on the water m ~ l e c u l e . ~Other theoretical calculations on HF have been made by Mueller,6 who used a limited MO treatment as well as an atomic orbital method and a method of semi-localized orbitals. An elaborate valency bond treatment of HF has been published recently by Kastler.6 Method of Calculation.-HF has a closed-shell ground state, with five doubly occupied MOs, and the method described by Roothaan' may be applied directly. The MOs are expressed as linear combinations of atomic orbitals

8.7;

CI

CI

= 2.6

The orbital ( 2 s ) ~is not used directly in the calculation; only the orthogonalized form s is used. The normalized ten-electron wave function of the normal state is an antisymmetrized product of the functions (1). The requirement that the energy corresponding to this state function shall be an absolute minimum leads t o the matrix equations (H G)ai = &ai (3) H, G and S in the present case are 6 X 6 matrices, while the a i are vectors, which are column matrices of the coefficients of the atomic orbitals in the +i of equation 1. The matrices H, G and S are made of elements of the operators H , G and unity, evaluated over the atomic orbitals (2). The definitions of these operators have been given in detai1,'J and will not be repeated here. T h e equation 3 is to be satisfied simultaneously by a set of (six) a i and corresponding eigenvalues t i . An iterative method of solution is required, since the elements of G depend on the a i . Molecular symmetry (Cmvhere) is used to reduce the matrix equation 3 into one set of four equations for the A1 orbitals and into two identical equations for the E orbitals. The molecular orbitals of proper symmetry are

+

(A1 Orbitals) = adz +Z = azlh +a = aaih 46 = asih +I

+ alzf + alas + a148 + + a235 + a242 + uszf + aa3~+ aa4z + + a& f u22f

Uezf

Us42

(+& is not occupied in the ground state)

(E orbitals)

The xp are specifically f,h, s,

2,

x, y, defined as

(1) Acknowledgment is made to the Office of Ordnance Research, U. S. Army, which supported this work in part under Contract DA-30115-ORD-295 with the University of Rochester. (2) A. B. F. Duncan and J. A. Pople, Trans. Faraday Soc., 49, 217

(1953). (3) J. E. Lennard-Jones, Proc. Roy. SOC. (London), 8198, 14 (1949). (4) F. 0.Ellison and H. Shull, J . Chcm. Phys., Pi, 1420 (1953). (5) C. R.Mueller, ibid., 19, 1498 (1951). (6) D.Kastler, J. chim. phys., 60, 556 (1953). (7) C. C. J. Roothaan, Rcu. Mod. Phys., 23, 69 (1951). A recent treatment of the CO molecule by this method has been made by R. C. Sahni, Trans. Faraday SOC.,49, 1246 (1953).

44 = x 46 = y

The secular equations for the A I orbitals were solved for the orbital energies and coefficients by the method of James and C o ~ l i d g e . ~ The final values of the coefficients are shown in equations 5 , where numbers in parentheses show values of coefficients introduced before the final solution. The orbital energies are given in Table I. The coefficients in (5) and corresponding +i are referred (8) J. F. Mulligan, J . Chcm. Phys.. 19,347 (1951). (9) H.M.James and A. S. Coolidge, ibid., 1, 825 (1933).

A. B. F. DUNCAN

2108

to as exact solutions. Actually the equations were solved first with the simplifying assumption of no inner-outer shell mixing, with ~3~ = 1 and a12, aZ2, a33 and a34 = 0. These solutions are referred to as approximate. The approximate functions +ia derived on this assumption are shown in equations 6 and the corresponding eigenvalues are shown in Table I.

+

41 = 0.3850h - 0.03503s 0.7400s f 0.10482 (0.3852) (0.03504) (0.7398) (0.1051) 42 = 0.1743h - 0.005702f - 0.3588s 0.89802 (0.1734) (0.005652) (0.3586) (0.8983) 41 = - 0.004916h 1.00012f 0.01877s 0.002946~’ (0.004919) (1.00012) (0.01877) (0.002945) 46 - 1 1326h 0.05271.f $. 0.8060s 3- 0.55982 1

1

+

1 1

+ + + + 4ia = 0.4054h + 0.7246s + 0.097222 (0.4055) (0.7244) (0.09746)

(5)

42’ = 0.1779h - ‘0.3644; +‘0.8953~.

(0.1771)

(0.3641)

=f 4 ~ ‘= -1.1221h 938

(0.8956)

(6)

+ 0.8168s + 0.56502

TABLE I CALCULATED ENERGY VALUES (in units ez/lao = 27.205 e.v., except where noted) Orbital energies ei (of +i)

- 1,5688 -0.5396 -25.9859 - 0.4156 1.1822 -28.9255 - 104.9037

El

e Ea €4

=

€6

(€1

€6

+

h

€3

Approxi mat e values

Exact values

- 1.6066 -0.5576 -26.0171 - 0.4334 +1.1032 -29.0481 - 104,9557

+

+ 2€4)

Total electronic energy ( E ) Nuclear repulsion energy

(En)

S5.1939 +5.1939 -99.7098 -99.7618 - 99.4700 - 99.4700 0.2398 0.2918 (6.52 e.v.1 (7.94 e.v.1 ‘0.2446 e2/ao S6.66 e.v.

Total molecular energy Energy of separated atoms

De De (exptl.)

The self-consistency of the results is seen better by the fact that the final total molecular energy is only about 0.001% lower than in the preceding trial; the corresponding increase in dissociatioii energy is about 0 . 0 5 ~ o . Total Energy and Dissociation Energy.-The total energy of the ground state is (in atomic units) &,tal

E

=

+ E , J ;E,,

=

ZpZH!rc=-F)

where E is the total energy of the electrons. E may be expressed aslo E

=

H,

2 i

The orbital energies

~i

+

(2Jl>-

(7)

Kij)

ij

satisfy the equation

from which E =

Hi

i

The values of

+

ti

i

4,

= 1

H+i dv

+

€1

i

CEi, E , and En are shown in Table i

I for the exact and approximate solutions. The (10) See ref. 6, equations 20-26, 31, and unnumbered equations between equations 64 and 65.

VOl. 77

energy of the F atom was calculated with the same atomic orbitals which were used in the molecular calculation. The energy of the H atom was added to get the energy of the separated atoms, each in its normal state. The difference between this sum and the total molecular energy is the dissociation energy, De, which is found to be 6.52 e.v. in the exact treatment and 7.94 e.v. in the approximate treatment. The experimental result is D e = 6.66 e.v., which is obtained by addition of the zero point energy to Doo = 6.40 e.v.*I The effect of inner-outer shell mixing is shown clearly in the result that the total energy is lower in the approximate treatment than in the exact treatment. In the exact treatment the calculated dissociation energy is better than might be expected. In the approximate treatment, however, the calculated dissociation energy is somewhat higher than the experimental value. Two comments may be made on this unexpected result. While the functions ( 5 ) obtained from the exact treatment are strictly orthogonal, there is an appreciable lack of orthogonality between +an and the other functions (6). It may be noted also that the calculated energy of the F atom ( - 9S.9700 e2/ao) is appreciably higher than an experimental value (-99.76 e2/ao) obtained from successive ionization potentials.12 The experimental value of the energy of the separated atoms is thus -100.26 e2/ao,or 0.79 atomic unit lower than was calculated. Possibly the reason for this lies not so much in the atomic functions themselves, but in the large Coulomb repulsion of the electrons. X similar situation must exist in the molecular case, which leads to the conclusion that the total molecular energy should be somewhat lower than has been calculated, both in the exact and a p proximate treatments. I t is difficult to estimate the relative repulsion in the atomic and molecular cases and the over-all effect on the dissociation energy. Since the latter is only about 0.4% of the total molecular energy, perhaps only an agreement in order of magnitude is to be expected for the dissociation energy. Ionization and Excitation Energies in HF.The negatives of the orbital energies form a good approximation to the ionization potentials of the inolecule.’ Unfortunately there are no known experimental values for HF. Theoretically, the lowest potential is expected (and found) to be that of $4 (or 2p,) = 11.3 e.v. This is considerably lower than the corresponding potential in the F atom (17.42 e.v.).l2 But it is questionable whether such a comparison should be made. The repulsions of the H atom and perhaps an excessive repulsion by the other electrons in x and y orbitals contribute large positive values to the matrix element which determines the orbital energy of x. This matrix element, and consequently the potential, is smaller than it would be in the free atom. The other calculated ionization potentials appear to have reasonable values. (11) G. Herzberg, “Molecular Structure and Spectra. I. Spectra of Diatomic Molecules,” D. Van Nostrand Co., Inc., New York, N. Y . , 1950, p . 5138. (12) C. E. Moore, “Atomic Energy Levels,’’ Vol. I, N a t . Bur. Standards, Circ. 467, U. S. Department of Commerce, Washingtan, 1849.

THEELECTRONIC STRUCTURE OF HYDROGEN FLUORIDE

April 20, 1955

The lower energy electronic transitions in H F result probably from excitation of an electron from a 44 (x) orbital, possibly to ns and nd atom like orbitals. Excitation to the antibonding orbital 46 can occur evidently only a t very high energies. The experimental data from ultraviolet absorption spectra are meager, are extremely difficult to obtain, and are complicated by association a t the pressures and temperatures necessary to obtain adequate absorption. l 3 Absorption appears to begin a t about 55000 cm.-' (6.0 e.v.) with indication of a maximum a t 62000 cm.-' (7.69 e.v.) a t 100-165". If this absorption corresponds to an orbital excitation 2px -+ 3s) then the term value of 3s in HF is about 40450 cm-l. The term value of 3s in the F atom is a t about 36000 cm.-'. Dipole Moment and Electron Distribution.-The total dipole moment M (in Debye units (D))is 5

M

&r+i dv

= -2eao i=l

+ ea0 n

Znrn

where r and rn are, respectively, the vector distances in atomic units of an electron and a nucleus from the origin of a suitable coordinate system. Here the origin is taken a t the F atom and the positive z direction extends along the internuclear axis or bond line toward the H atom. The total moment will have a positive resultant value in the positive z direction; in this direction the individual moments in the first summation may have positive or negative signs, depending on the wave functions. The resultant dipole moment M calculated with the set of functions (5) is 2.65 D and with the approximate functions (6) is 2.53 D. Consequently there is no improvement in the calculation by the inclusion of ( 1 s ) ~mixing. This is in contrast to the results of Ellison and Shul14 on the water molecule. The calculated values are about 33-39% higher than the experimental value (1.91 D). It may be pointed out, however, that this calculation is extremely sensitive to the coefficients in the functions used. The values are about equal to values reported by KastlerP6but are inferior to Mueller's value from semi-localized orbitals. An analysis of the contributions of the individual moments of the different electron distributions to the total moment is shown in Table 11, together with the moment arising from the nuclei, It is apparent that the moment of the distribution 141l2 is negative in the direction of the total moment, as would be expected from the bonding character of this molecular orbital. The moment of the distribution 142 l a has the opposite sign, which indicates that this distribution is directed away somewhat from the bond direction. The functions 41 and 42 and qha)are not sufficiently localized, however, to be described, respectively, as bonding and lone pair functions. But if these molecular orbitals are subjected to the orthogonal transformation $1'

=

$2'

=

42 cos X -+2 sin

+

+l

a+

sin h cos

x

(8)

and if X is adjusted to make a21 vanish, there is obtained the following sets of MOs (13) E. Safari. Ann. phys., Series 12, 9, 203 (1954).

+ +

2109

+

- 0.3426s 0.5262s 0.46582 42' = -0.009253s 0.6321s 0.77482 4ia' = 0.4427h 0.51705 0.44882 $2"' = 0.62495 - 0.78082

41' 0.4226h

+

+

(9) (10)

These sets show more clearly the distribution of electrons into lone pair and bonding molecular orbitals. The moments of electron distributions corresponding to the functions (9) and (10) are quite different from the moments obtained with the functions (5) and (6). The moments are shown in Table 11. It is clear that cp'z and +a'z give quite appreciable lone pair moments. Of course neither the total resultant dipole moment nor the molecular energy is changed by this orthogonal transformation. TABLE I1 DIPOLEMOMENTS (IN DEBYEUNITS) Moments ( - 2 eao J $i z +i da) ( z = IF cos BF; f+p = angle between electron distance from F atom and internuclear axis) Function

91 92

9s

- H)

F% (t& '?

M (total)

Exact value

Approximate value

-2.7198 $0.9645 -0.0015 $4.4042 f2.6474

- 2.8426 +O. 9660 Zero +4.4042 f2.5276

Moments from transformed functions (9), (10) -4.6526 -4.5484 $1, +2.7830 $2,7758 +2' Zero $8' = 93 -0.0015 +4.4042 eUoZHr(F H) +4.4042 +2.5274 M (total) +2.6473

-

An equivalent orbital representation of the lone pair electrons can be obtained from +'2 (or 4a'~), 44 and 46 by a further orthonormal transformation. Let xll, XI,, xla be the equivalent orbital functions which have the forms and properties described in the first paragraph (see also reference 1 for further details), Then

and, for example XI, =

-0.00534j

xllS = 0.3608s

+ 0.36495 - 0.44742 + 0.8165%

- 0.45088 + 0.8165s

(12)

etc., may be obtained also from equations (9, 10, 11). The moment of the distribution (I xil 1' I XI, l 2 I xla12) is, of course identical with the moment of the distribution I +'2 12. However, the forms of the equivalent functions (12) enable one to estimate roughly the angle between lone pairs, and to deduce additional features of the electron distribution. In the notation of ref. (2) xtl = cos eis sin el cos y z sin 81 sin y x where y is the angle between XI: and the internuclear axis. Using xllain equation (12)

XIZ,

+

+ +

+

2110

IVAR T. KROIIK,It. C. M-ERPL'ER.ZND HYMINSIIAPIRO

Vol. 77

sins1 cos y = -0.4508; sin el sin y = 0.8165 el = 0.93268; sin y = 0.87544; cos y -0.48334.

Sources of Integrals.-All integrals over atomic orbitals may be evaluated exactly except twocenter exchange integrals, and even for these, five If p is the angle between lone pairs, since or six terms of the infinite series are sufficient. sin (P/z) = (&/2) sin y Formulas suitable for direct numerical evaluation then p = 98'36'. are available for most integrals. A large number In an equivalent orbital representation, there of integrals have been evaluated and tabulated by is a single bonding function for HF, which is just Kotani, et al., l5 but uncertainties in interpolation 9'1 (or 41~'). The approximate function may be made recalculation preferable in the present work. written in the form These tables, however, served as a valuable check. most of the required integrals had been After '$1" = XB' = p h (COS E @ + Sin EbZ) (13) calculated, the results and tables of integrals for and from (10) it is found that X = 0.68469, ( X / p ) = H F of KastlerlG became available, and provided 1.55, COS Eb = 0.75523 and sin Eb = 0.65547. The polarity (H+F-) is much higher, as measured by additional checks on the integrals. The one-centered integrals were evaluated from (X/p), than the polarity of a function made to give the formulas of Coulson,17 Barnett and Coulsonl* the correct dipole moment. For example, from and Roothaan'g ; some integrals were computed the previous work,2such a function with p = 98.b0, by several methods. Two-centered integrals were will have (X/p) about equal to 1-09. If the angle p between lone pairs is accepted as computed a t the normal H-F distance, 1.7328~0." Two-centered Coulomb integrals were obtained about 98.5', it is of interest to compute the position from the formulas of R ~ o t h a a n . ' ~A11 hybrid of the ring of maximum electron density in the lone integrals were evaluated by the analytical methods pair region of the molecule. This calculation is of Barnett and Coulson. Two-centered exchange interesting in connection with association of H F to integrals were computed by the methods of Ruedenform polymers. Electron diffraction experiments14 bergz0; nunierical integrations were used for his indicate that the principal associated form has a &,-functions, while his Bj functions were interzig-zag structure in which the average F-F-F polated from the tables of Kotani.15 Tables of all angle is 140 f so, and the H atoms are on line integrals used in these calculations are available.21 with the F atoms to about +15". I t might be (1.5) b l . Kotani, A . Amemiya and T. Simose, Proc. Phys.-hfalh. Soc. expected that the H atom of one molecule would p a n , 20, Extra No, 1 (1938); 22, Extra No. 1 (1940). attach itself a t the position of maximum electron J o (16) The integrals are tabulated in ref. 5 . T h e author is verv density in the lone pair region of another molecule, grateful t o Dr. Kastler for furnishing him a copy of reference 5 before and that the polymer is formed by repetition of its publication. (17) C A . Coulson, Proc. Cambridge Phil. Soc., 38, 210 (1941). this process. The calculation of the angle (18) M. P. Barnett and C. A . Coulson, Phil. Trans. R o y . Soc. (Lonwhich the ring of maximum density makes with the don), 8243, 221 (1951). bond axis was made in the former paper2 for sev(IY) C. C . J . Roothaan, J . Chein. Phys., 19, 1445 (1'351). (20) K . 121-edenherg, ibid., 19, 1469 (1951). eral values of the angle 0. The requisite formulas (21) Tables of integrals have been deposited as Document 4451 are given in that paper and will not be repeated with t h e AD1 Auxiliary I'ublication Project, Photoduplication Service, here. For /3 = 08.5", with the values of the pa- Library of Congress, \Vashin~ton,D. C. A copy may be secured by rameters given here from (12) and (13), it is found citing t h e Dccument number and by remitting $1.25 for photoprints or $1.25 for 3R mm. microfilm in advance by check or money order that Omax, is about 141'. sin

+

(14) S. € I . Bauer, J. 19 (1930).

Y .Beach and J . H. Simons, THISJ O U R N A L ,

payable t o : Chief, Photoduplication Service, Library nf Congress.

61,

ROCHESTER, S . Y. [CONTRIBUTION PROM THE

RESEARCH LABORATORIES OF THE ETHYL CORPORATION]

The Compound NapPb, BY IVART . K R O H N ,R. ' ~ C. W E R N E RAND ' ~ HYMINSHAPIRO RECEIVED SEPTEMBER 22, 1954 A little-known region of the Na-Pb equilibrium diagram, in the vicinity of the composition KaePbr, was reinvestigated by means of thermal and microscopic analysis. The work demonstrated the existence of a hitherto unrecognized openmaximum compound at the NaoPb, composition. The neighboring compound NasPbz was shown to be a peritectic rather than an open-maximum compound.

Previous investigations2 of the complicated NaPb equilibrium system have indicated the existence of five compounds: NaaPb6,NaPb, NanPb, NaJ& and NaaPb. I n addition, although neither cornpound is formed from melts, Na4Pb, and N&Pb9 (1) (a) T h e Haloid Corporation, Rochester, N. Y. (b) Mine Safety Ai>i>liancesCo., Pittsburgh 8, Pa. ( 2 ) M. Hansen, "Der Aufbau Zweistofflegierungen," Julius Springer, Berlin, 1936.

have been reported in liquid ammonia ~ o l u t i o n . ~ The compound NazPb4 was considered to be formed a t 182" by the solid-solid reaction NabPbz(s) NaPb(s) = 31L'azPb(s). The other compounds in the Phase diagram were reported to be of the openmaximum type, with Na6PbZgiving the highest

+

( 3 ) E. Zintl, J. GouLeait and W. I>iillenkuyf, 2 physik. C h e m . , 8 1 6 4 , 37 (1Y31). (4) G . Calingaert and W. .T Boesch, Trris JOURNAL, 46, 1901 (1923)